Distributed agent-based detection system using PSO and neural network for MANET

Abstract

Due to dynamic and infrastructure less nature of mobile adhoc network, it is prone to various kinds of attacks. In this paper, we have proposed a distributed agent-based detection system for network layer attacks using PSO and neural network in MANET. In the proposed technique four agents are considered with different complementary role in order to detect attack. For efficient detection of the intrusion, the network is divided into small zones. In each zone one detection agent is presents to manage and detect the irregularity in node activities. Detection agent is found by using particle swarm optimisation PSO technique. After the detection of invader, response agent is generated and is reacted with the invader to avoid future damage. Also, back propagation neural BPN network is collaborated with the detection agent to detect the network attacks.